TimesFM Forecasting

What it does

• Runs zero-shot forecasting on any univariate series using Google Research's TimesFM 2.5 (200M params) — no model training needed.
• Returns both a point (median) forecast and 10 quantile bands (q10–q90) so you can build 80%/90% prediction intervals.
• Accepts CSV, pandas DataFrame, or NumPy array inputs and supports batch forecasting of hundreds or thousands of series.
• Ships a mandatory preflight checker (scripts/check_system.py) that validates RAM, GPU VRAM, disk space, and Python version before the ~800 MB weights are downloaded.
• Documents covariate forecasting (forecast_with_covariates()), quantile-based anomaly detection, ARIMA baseline comparison, and 10 common pitfalls.

Who it's for

• Analysts and data scientists forecasting sales, demand, sensors, energy, vitals, or weather.
• Teams who want a fast foundation-model baseline instead of hand-tuning ARIMA/ETS.
• Inventory and capacity planners who need uncertainty bands, not just a single number.
• Not a fit if you need interpretable statistical coefficients, series classification/clustering, or multivariate causality analysis.

Example uses

1. 12-month sales forecast: load monthly_sales.csv, convert each column into a 1-D array, call forecast(horizon=12), and emit a table with 80% lower/upper bounds.
2. Sensor anomaly alerts: flag actuals outside the 80% interval as warnings and outside the 90% interval as critical.
3. Promo-aware demand forecast: pass price and holiday covariates via forecast_with_covariates() and compare MAE against plain TimesFM and an ARIMA baseline.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/K-Dense-AI/scientific-agent-skills/HEAD/skills/timesfm-forecasting/SKILL.md

What I want: (describe your task here)

Install the skill I found on AI Skill Mart.
Copy the skills/timesfm-forecasting folder from the GitHub repo K-Dense-AI/scientific-agent-skills into my ~/.claude/skills/timesfm-forecasting/.
When it's done, tell me in one line what this skill can do.